GB2137774A - Automatic Control of Friction and Inertia Welding Process - Google Patents

Abstract

Inertia welding and friction welding processes are controlled by predetermining the ideal rates of relative rotation, force applied and metal upset rates for a given material to be welded, enters those rates in a microprocessor memory and then connects the microprocessor to the apparatus on which the weld is to be effected. The microprocessor 24 monitors the actual rates, compares them with the ideal rates and should differences occur, generates signals from those differences, with which to adjust operation of the apparatus by means of ram 28. <IMAGE>

Description

SPECIFICATION Improvements in or Relating to Control of Friction and Inertia Welding Process Inertia welding is a known process for joining two metallic components and comprises fastening one component to and coaxially with a flywheel, driving the flywheel up to a given speed of revolution, thereby generating a given value of energy, forcing the rotating component against the static component after removing the drive from the flywheel. The energy stored in the flywheel continues to rotate the component and the resultant friction between the relatively rotating components in turn generates sufficient heat, in the first instance to soften the interface, and secondly to assist the pressure applied to achieve a solid phase weld.

Friction welding is a further known process for welding and differs from inertia welding in that the rotary drive is powered, rather than being derived from stored energy and the rotation may be stopped by means of a brake rather than the dissipation of the stored energy.

During both processes, metal is displaced or "upset", which results in a shortening in the overall axial length of the two components. The total amount by which the length of the two components is reduced may be termed the "upset length". In friction welding, at least, this can be subdivided into an initial "burn-off" or "friction" length (a reduction in length caused during a first period of time when heat is being generated to soften the interface) and a "forge length" (a subsequent reduction in length while the applied pressure forges the two components together).

Commonly, the applied pressure is increased after the burn-off length has been achieved, so that the forging pressure is greater than the friction pressure. In inertia welding, there is generally less distinction between these two phases of the welding process. A discussion of such matters may be found in British Standard BS 6223: 1982, published by the British Standards Institution.

U.K. Patent GB 1254022 is an example of friction welding apparatus in which the speed of the drive motor is controlled in accordance with a pre-programmed speed-time relationship. Such techniques are useful for controlling the quality of the resulting weld. However, such techniques give no control over the resulting amount of upset metal, and in particular give no control over the upset length. For example, in friction welding of two typical components by a conventional method, there might be a tolerance of +0.5 mm in the upset length. Moreover, speed control as shown in GB 1254022 is not possible in inertia welding, since it is impossible closely to control the speed of the massive flywheel used, owing to its inertia.In inertia welding of two typical components by a conventional method, since there is no control of the initial burn-off length, there might be a tolerance of +1 mm in the upset length.

Because there is such a tolerance in the upset length, critical components need to be designed oversize, and a large amount of expensive machining to size is needed after the welding operation. Furthermore, in some applications such as disc-to-disc welding of rotor disc assemblies in gas turbine engines, subsequent machining to size may be impractical or impossible because of the complicated shapes of the components.

Accordingly, conventional inertia and friction welding processes are not feasible for such applications.

U.K. Patent GB 1293141 relates to a method of controlling friction weld quality, and states that the rate of the initial burn-off is of more importance than the actual burn-off length.

Accordingly, this patent proposes monitoring the burn-off rate (the rate of change of length will respect to time during the initial burn-off period).

The burn-off rate is compared with a pre-set reference value and the axial welding pressure is adjusted to keep the burn-off characteristic following a desired straight line. This method cannot control the resulting overall upset length, for two reasons. Firstly, only the rate of metal upset is controlled, for weld quality reasons. This gives no direct control over upset length, and the patent is not concerned with controlling upset length. Secondly, the control is only during the burn-off phase, and there is no suggestion of control during the forging phase, which has an important effect on total upset length.

Another inertia welding process is shown in U.K. Patent GB 1439277. In this patent, in order to provide assurance of the quality of the weld, the pressure, speed and upset are monitored throughout the-welding process. Should the pressure, speed or upset stray outside predetermined ranges within which a good quality weld can be assured, an indication of this fact is provided, and the welding process may be automatically stopped. There is no feedback control over the welding parameters. Such a quality control method is only acceptable because in practice the predetermined ranges within which the parameterscan vary while still producing a good quality weld are quite wide.

Were this not so, a large proportion of the components welded would need to be rejected.

For example, ranges of ~7% permissible variations in the nominal, "ideal" weld parameters would be common, The problem with which the present invention is concerned, however, is control of the total upset length in situations where such variation would be intolerable.

According to the present invention, there is provided a method for controlling a process in which two components are rotated relative to one another and pressed together, so as to be welded together by inertia or friction welding, comprising monitoring the instantaneous value of the upset length of the components at a plurality of predetermined points during the progress of formation of the weld, comparing each monitored value of the upset length with a predetermined value of upset length at the corresponding point of an ideal welding operation, deriving a signal from any difference therebetween, and utilising that signal to change the pressure applied between the components so as to at least reduce any said difference.

Preferably, where the welding process is an inertia welding process, the force applied is controlled in accordance with signals derived from differences, in actual and ideal magnitude of metal upset relative to a given rate of rotation.

Preferably, where the welding process is a friction welding process, the force applied is controlled in accordance with signals derived from differences, if any, in actual and ideal magnitude of metal upset relative to time.

The invention will now be described by way of example and with reference to the accompanying drawings in which: Figure 1 is a schematic view of apparatus utilised in the operation of the processes in accordance with the invention.

Figure 2 is a graph of speed of relative rotation of the components, ideal rate of metal upset and force applied plotted against a given time in accordance with one aspect of the present invention relating to inertia welding, Figure 3 is a flow chart of a program for a micro-processor unit, for use in operation of the process depicted in Figure 2, and, Figure 4 is a graph similar to Figure 2 except that it relates to friction welding.

In Figure 1 a flywheel driven chuck 10 supports a first component 12 for rotation. A further component 14 is rigidly held in a vice 1 6 which in turn forms part of a hydraulic ram mechanism 1 8, the function of which is to enable vice 1 6 and therefor component 14 to move towards and away from component 12.

Ram mechanism 18 is connected to a fluid pressure system consisting of oil supply 20, a pump 21, and a servo valve 22.

A microprocessor unit 24 has a memory 25 and a control console 27. It is connected to servo valve 22 for the purpose of controlling the oil pressure to ram mechanism 18. The microprocessor 24 is further connected to receive signals from a pressure transducer 26 by means of which the microprocessor 24 monitors the oil pressure in the cylinder 28 of ram mechanism 18.

A linear transducer 30 which indicates the rate and magnitude of relative movement of components 12 and 14 towards each other and a tachometer 32 which indicates the speed of rotation of the chuck 10, and, therefore, the speed of relative rotation of components 12 and 14, are also connected to the microprocessor 24.

Referring now to Figure 2. Prior to operation of the apparatus to weld together two given components on a production basis, the ideal values of metal upset throughout the welding operation are determined along with the ideal values corresponding thereto of the relative rotation between the two components 12 and 14 (Figure 1) at a nominal, given applied force through ram mechanism 18 (Figure 1). This is done in a manner known to those skilled in this art, by means of test welds on sample components of the same size, shape and material as the production components.

Ideal upset values are depicted by the full line 34, and the ideal rate of fall off of relative rotation is depicted by full line 36. The data so obtained is stored in microprocessor memory 25, in the form of a table of values of relative rotation and a table of the corresponding ideal upset values. The nominal applied force is depicted by full line 38 and is bounded by positive and negative value limit lines 40 and 42 respectively, which might be, for example, +10% of the nominal value 38.

These upper and lower limit values are also stored in the memory 25.

At the start of the operation, components 12 and 14 are separated and component 12 is brought up to a predetermined ideal speed of revolution via the power driven flywheel (not shown), the speed of revolution being indicated at point R in Figure 2. At this point, drive is disconnected so that the flywheel, chuck 10 and component 12 commence free wheeling and at the same time, components 12 and 14 are brought together under the action of ram mechanism 18. The force applied thus is the nominal force indicated at point P.

The initial meeting of components 12 and 14 (Figure 1) under the conditions described herein generates friction, reduces the rate of rotation of component 1 2 and effects softening of the components at their interface. Consequently the force applied moves component 14 further towards component 12, which movement is sensed by linear transducer 30 and is passed as a signal to the microprocessor 24 where it is compared with the value corresponding to the appropriate point on line 34 of Figure 2.

The microprocessor 24 will monitor the force applied via pressure transducer 26 and so manipulate the servo value 22 so as to ensure an appropriate oil delivery to ram mechanism 18. If, however, the actual value of metal upset changes relative to the value depicted by line 34, for example, by way of an increase as indicated by dotted line 44, that change will immediately become apparent by virtue of the microprocessor 24 comparing the signal from the linear transducer 30 with that contained in its memory.

The increase in the value of metal upset at specific values of rate of rotation produces, via the microprocessor 24, changes in the signal to the servo valve 22 which in turn adjusts the oil pressure and hence the force applied through the ram mechanism 18. In the specific example described, when the rate of rotation has fallen to point R1 on line 36 the value of metal upset at point U1 on- line 44 is seen to be high relative to ideal value of metal upset on line 34.

The microprocessor 24 compares the actual value on dotted line 44 with ideal value on line 34 and manipulates the servo valve 22 to decrease the pressure from P1 by predetermined discrete steps; When the rate of rotation has further fallen to point R2 on line 36, the actual metal upset U2 once again is as indicated on line 34 and the pressure has fallen to pressure P2. The microprocessor 24 then manipulates the servo valve 22 to maintain the pressure at P2. Again when the rate of rotation has further fallen to point R3 on line 36 the value of actual upset U3 on line 48 is seen to be low relative to ideal value of metal upset on line 34. The microprocessor 24 compares the actual value U3 on line 48 with the ideal value on line 34 and manipulates the servo valve 22 to increase the pressure from P3, again by predetermined discrete steps.When the rate of rotation of actual metal upset U4 is again consistent with the ideal value on line 34 and the pressure has increased to pressure P4, the microprocessor 24 manipulates the servo valve 22 to maintain the pressure at P4.

The force applied, however, should never be varied such that it rises or falls beyond the values represented by lines 40 and 42 respectively.

Figure 3 is a simplified flowchart of the program followed by the microprocessor in order to achieve the above. In the first stage 70 of the program, the microprocessor initializes itself, and waits for the commencement of a welding cycle.

At this stage, pointers to the tables of relative speed of rotation and upset are initialised. Once a welding cycle commences, these table pointers are incremented (stage 72) so as to point to the first values in the tables, and a reading is then taken from the tachometer 32 (stage 74). At decision stage 76, the microprocessor tests whether the speed of rotation (RPM) has fallen to the current value in the table; if it has not, then the program loops back until this test succeeds.

The value of upset which has been achieved is now read from the linear transducer 30 (stage 78).

It is only desired to control the pressure while the weld is actually progressing, and before and after this control region of the process, the microprocessor can be programmed simply to monitor and record the various parameters (by program stages not shown in Figure 3). So the microprocessor next tests (at stage 80) whether the process is in this control region and loops back if not.

If control is required, the actual upset value read at stage 78 is compared (at stage 82) with the ideal upset value stored at the current location in the table. It will be appreciated that this ideal upset value is that which should have been achieved in an ideal welding operation at the current value of rotational speed, as indicated in Figure 2. Should the actual upset be higher than the ideal (test stage 84) and if the measured pressure is above the lower limit line 42 (test stage 86), then the servo valve 22 is adjusted to lower the pressure to the ram 1 8 by a predetermined amount (stage 88). Similarly, the pressure is raised by a predetermined amount (stage 94) if the actual upset is lower than the ideal (test stage 90) and the pressure is below the upper limit 40.In any event, the program loops back to stage 72, at which the table pointers are incremented, and the program then waits for the rotational speed to fall to the next monitor point, indicated by the new value in the RPM table thus pointed to.

The maintaining of precise control as described herein whilst metal upset is occuring, ensures that the ideal quantity of metal is displaced. The final overall length of the welded components 12 and 14 is thus more nearly predictable and obtainable.

This in turn obviates or at least reduces to a minimum any machining necessary to achieve that dimension. In practice, we have found it possible to control the total upset to within +5 thousands of an inch (+127 yam). This tolerance is thought to be governed by the equipment used rather than being a parameter of a particular welding operation or particular components, and is relatively independent of the actual value of the total upset. In test weldings of 2 inch (51 mm) diameter cylinders of mild steel, this tolerance was repeatably achieved in a total upset length of 0.187 inches (4.75 mm), which in this particular case is a tolerance of 2.7%. This compares most favourably with the tolerances of conventional methods.

Referring now to Figure 4. The graph depicts on its vertical X axis the values of relative rotation between components 12 and 14 of Figure 1, force applied and the magnitude of metal upset, plotted against the ideal times depicted on the horizontal Y axis, over which those values should be attained. The graph is relevant to a friction welding operation, i.e. a process wherein relative rotation of the components is achieved by continuously driving one of them.

Line 62 depicts ideal speed of relative rotation of components 12 and 14, line 64 depicts the ideal force, application of which commences at point F when the ideal rate of revolution has stabilised.

Line 66 represents the magnitude of metal upset to be achieved over a given time period depicted by line 68 which is coincident with the Y axis.

The ideal values for each function are entered into the memory of a microprocessor of the kind described in connection with Figure 1. Force applying means 1 8 of the kind described in Figure 1 is provided, as is a linear transducer 30, a pressure transducer 26 and a tachometer 32.

In the present example, the speed of revolution is achieved by powered means (not shown) and consequently is not affected by conditions at the interface. If, however, the value of metal upset varies relative to that represented by line 66 in Figure 3, i.e. the upset which should be achieved in a time represented by line 68, has not been achieved or has been exceeded the relevant signals will be entered in the microprocessor 24, compared, and an appropriate correcting signal issued by the microprocessor 24 to the servo valve 22 to the force applying means 18. Thus if dimension "A" which represents the desired quantity of metal to be displaced is being approached at a rate which indicates that it will be achieved before time "T" is reached the force being applied will be reduced on instruction from microprocessor 24 (Figure 1).Similarly, if dimension "B" which represents the desired magnitude of metal upset during the final joining stage of the process is being achieved in some time which is different from that indicated by Tt, the force applied will be varied appropriately, by signals generated in mocroprocessor 24 (Figure 1). The microprocessor thus ensures that the total magnitude of metal upset is limited to that indicated by dimension "C" and on that being achieved, a signal is generated which maintains a force for a time T2 before switching off of the apparatus. It will be seen that whereas the apparatus of Figure 1 compares the actual and ideal values of the upset at predetermined values of RPM, the present apparatus performs this comparison at predetermined times. The program to be followed may be generally similar to that of Figure 3, except that instead of waiting for the RPM to fall to the next monitor value, the program samples and compares the upset at the appropriate predetermined times throughout the welding operation.

While the above embodiments have been described as microprocessor-based units, it will be appreciated that other forms of data processing means can be used if desired.

Claims (6)

1. A method for controlling a process in which two components are rotated relative to one another and pressed together, so as to be welded together by inertia or friction welding, comprising monitoring the instanteous value of the upset length of the components at a plurality of predetermined points during the progress of formation of the weld, comparing each monitored value of the upset length with a predetermined value of upset length at the corresponding point of an ideal welding operation, deriving a signal from any difference therebetween, and utilising that signal to change the pressure applied between the components so as to at least reduce any said difference.

2. A method according to Claim 1 for controlling an inertia welding process, in which the upset length is monitored and compared at a plurality of points corresponding to predetermined values of the relative speed of rotation of the components.

3. A method according to Claim 1 for controlling a continuous drive friction welding process, in which the upset length is monitored and compared at a plurality of predetermined times during the welding operation.

4. A method according to Claim 1, Claim 2 or Claim 3 including determining the pressure applied between the components, and only utilising said signal to raise the pressure if the pressure is below a predetermined upper limit, and only utilising said signal to lower the pressure if the pressure is above a predetermined lower limit.

5. A method according to any one of the preceding claims wherein said predetermined values are stored in a memory of data processing means which is programmed to perform said steps of monitoring, comparing and deriving said signal.

6. A method for controlling a process in which two components are rotated relative to one another and pressed together, so as to be welded together by inertia or friction welding, substantially as any described herein with reference to the accompanying drawings.